Hi, when I delete a relationship from the Playground and import in back in, the relationship still appears?

Hello, is there a way to generate

RelationshipFilter

to match all relations by

SubjectType

&

SubjectID

, without passing (or using a wildcard) to required

ResourceType

field?

For a scenario where I want to bulk delete relationships with certain subject.

currently, no, you need to give at least the relation type as well

Okay, thank you

I think there is no way to unionise multiple filters too?

nope

we've had suggestions to allow for subject-only deletions in the past

Sorry for the delayed response but this looks correct

Question on the lead time: if I were to get SpiceDB dedicated, approximately how long after the contract is signed will the provisioned cluster be ready for development on our side?

It depends on the requirements (e.g. govcloud, etc..), but if you're in a standard AWS region, the turn around is a business day, maybe two at the worst.

does anyone happen to have a PEG parser definition for spicedb's schema language lying around that they wouldn't mind sharing? i've got some codegen ideas i want to try out and in the absence of official tools that seems like the quickest way to extract info about the schema

there isn't a formal definition, but the language is fairly simple (right now). what language are you writing the codegen in?

rust. for sure, i'm totally happy to put it together myself i just wanted to avoid repeating work if i could

ah; yeah, we have a golang-based parser and we do intend to have a reflection API shortly

the go entrypoint is here and documented on each form: https://github.com/authzed/spicedb/blob/main/pkg/schemadsl/parser/parser.go#L29

should be fairly easy build a rust parser or grammar off of it

awesome. thank you!

keep an eye on https://github.com/authzed/spicedb/issues/439 too

that's one of the two reflection APIs we'll be starting shortly

I'm still really early in understanding SpiceDB. I was wondering how people think about their application design, and "datastores" in general, that "live along side SpceDB". For example, we use DynamoDB as our primary datastore. I've read up on the ZedTokens. I think i get that part of it..... From an operational POV, it feels like the SpiceDB datastore, needs all the same "-isms" as my Application Datastore. Operational expertise, Backups, HA, multi-region, etc. Maybe some specific questions: - Do people write their "list of editors" for an object, also in their Application datastore? Or is it only stored in SpiceDB? eg, do you store enough data in your Application that you could "reconstruct" the relationships? - At some scale the requirements for spiceDB are probably different scaling characteristics than my primary application -- are people deploying them as parallel systems, or are people re-using their existing database and also running SpiceDB on it?

> Do people write their "list of editors" for an object, also in their Application datastore? Or is it only stored in SpiceDB? eg, do you store enough data in your Application that you could "reconstruct" the relationships? Usually application specific, but often times you can keep the relationships between objects in SpiceDB and the metadata for the objects in your application database. For example, members of groups in SpiceDB, but descriptions of groups in your application database. Some others will keep a copy in both. > - At some scale the requirements for spiceDB are probably different scaling characteristics than my primary application -- are people deploying them as parallel systems, or are people re-using their existing database and also running SpiceDB on it? Typically, it is recommended to run a distinct database since the traffic patterns will be different and, as you mentioned, the scaling requirements are very different. In our own experience, permissions checks can end up being a SIGNIFICANT part of the traffic to an application database before moving to a SpiceDB-like model, which means you might be able to reduce the scale of your application database

Have you seen people use CDC/ Change Data Capture in their Application database, to then do writes to SpiceDB? Wondering about "sync" issues really. Eg, if a write fails to SpiceDB... but my App has already saved an object.

yes and, as you suggested, there are cases where that can be an issue if you're not careful

it really depends on your consistency requirements

i mean, i'd pick zero of 3 if i had a choice, but customers, you know, they like Ponys and perfect consistency 🙂

hello , new here!

If i were modeling tags to documents to control what tags someone can add to docs... how do i represent in relation definition tag { relation : ? } definition doc {}

are tags unique for a document?

generally speaking, unless you have major consistency concerns, the recommendation is to either use CDC or open a db transaction, write to SpiceDB, then complete the DB transaction. See: https://authzed.com/blog/writing-relationships-to-spicedb/